Validating delay constructs: An application of confirmatory factor analysis

This paper proposes to use confirmatory factor analysis (CFA) to evaluate the relationship between six observed variables (arrival and departure counts, arrival and departure demand, taxi-out and airborne delays) and their underlying latent (unobserved) constructs (operations, demand, and delays) at six of the most delayed airports (EWR, JFK, LGA, MIA, ORD, and SFO) during the calendar years of 2006–2008. The CFA revealed a good fit between the six observed variables and the three factors that may explain on-time performance except in the case of JFK. The use of CFA can help analysts validate constructs when theory supports a priori predictions and relationships between observed and unobserved variables.